Novel energy effcient protocols with realistic radio propagation models for wireless sensor networks in agriculture

Abstract

A wireless sensor network (WSN) is an emerging technology that enables a variety of possible applications. The performance of a WSN depends on many factors such as the physical layer parameters (i.e. transmission power and frequency selection), and the communication protocols (MAC and routing protocols). Accounting for these factors, the technical challenges remain in designing and deploying a robust WSN with limited energy supplies in a harsh environmental condition, where the unreliability of wireless propagation channel restricts the performance of the sensor node. Researchers have invested a lot of time and e ort into developing high performance communication protocols to meet the growing challenges of WSN. However, there is still no approach that is able to meet the requirements and challenges of agriculture application, especially in realistic simulation design of WSN protocols. This thesis focuses on the simulation of the proposed novel energy efficient protocols that are robust to variations in the radio propagation environment. The proposed protocols ensure the connectivity between the cluster members and cluster head (CH), applicable for dense networks, low network overhead and most importantly, energy e cient. To address this, an actual measurement on vegetation attenuation is carried out to ascertain the best propagation model for WSN protocol design and evaluation in a simulation platform. Based on these models, a MAC layer algorithm designed for clustering protocols such as LEACH, called AgriMAC is proposed, and it is combined with dynamic transmit power control algorithm, DytCon. The performances of these algorithms are compared with LEACH in term of energy e ciency and network lifetime. Results show that the performance of these algorithms achieves a substantial improvement in network lifetime compared to LEACH. AgriMAC eliminates the penalty of having more cluster heads to the network lifetime, where a steady performance is achieved when the number of cluster is between 4 and 10 with approximately 4:8% improvements over the maximum net- work lifetime achieves by LEACH. In order to solve various issues of LEACH clustering protocol such as unbalanced network partitioning and variable number of cluster heads per round, a novel energy eficient clustering protocol, DynClust is proposed. The protocol combines a machine learning technique called k-means, where it groups the nodes into clusters with AgriMAC and DytCon to optimize the network lifetime. DynClust exhibits vital properties such as robustness against variations in the radio propagation environment, a very low control overhead, simple and yet e cient. The protocol improves LEACH in term of cluster distribution and cluster membership. From the simulation results, DynClust achieves approximately 318% improvements over LEACH in term of network lifetime in various prop- agation environments. These allow the possibility of WSN to be simulated accurately in dynamic and harsh agriculture applications.